Publisher’s version / Version de l'éditeur:
Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez pas à les repérer, communiquez avec nous à PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca.
Questions? Contact the NRC Publications Archive team at
PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the first page of the publication for their contact information.
https://publications-cnrc.canada.ca/fra/droits
L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.
Biochemistry, 34, 20, pp. 6737-6744, 1995-05-23
READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE.
https://nrc-publications.canada.ca/eng/copyright
NRC Publications Archive Record / Notice des Archives des publications du CNRC :
https://nrc-publications.canada.ca/eng/view/object/?id=1df6fea1-59aa-4dbc-90d2-3463988cd415
https://publications-cnrc.canada.ca/fra/voir/objet/?id=1df6fea1-59aa-4dbc-90d2-3463988cd415
NRC Publications Archive
Archives des publications du CNRC
This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. / La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version acceptée du manuscrit ou la version de l’éditeur.
For the publisher’s version, please access the DOI link below./ Pour consulter la version de l’éditeur, utilisez le lien DOI ci-dessous.
https://doi.org/10.1021/bi00020a019
Access and use of this website and the material on it are subject to the Terms and Conditions set forth at
Evidence for the extended helical nature of polysaccharide epitopes.
The 2.8 A resolution structure and thermodynamics of ligand binding of
an antigen binding fragment specific for alpha-(2-->8)-polysialic acid
Evans, S.; Sigurskjold, B. W.; Jennings, H. J.; Brisson, J.-R.; To, R.; Tse, W.
C.; Altman, Eleonora; Frosch, M.; Weisgerber, C.; Kratzin, H. D.; Klebert, S.;
Vaesen, M.; Bitter-Suermann, D.; Rose, D. R.; Young, N. M.; Bundle, D. R.
Evidence
for
the
Extended
Helical
Nature
of
Polysaccharide Epitopes. The
2.8
A
Resolution Structure
and
Thermodynamics
of
Ligand
Binding
of
an
Antigen
Binding
Fragment
Specific
for
a-(2—1'8)-Polysialic
Acid*
*S. V. Evans,*°° v
B. W. Sigurskjold,v§ H. J.
Jennings,*v J.-R. Brisson,vR. To,v W. C. Tse,v E. Altman,v
M.
Frosch,11 C. Weisgerber,v H. D. Kratzin,A S. Klebert,A M. Vaesen,A D. Bitter-Suermann,11 D. R.Rose,vl
N. M. Young,v andD. R. Bundlev’#
Institute
for
BiologicalSciences, NationalResearchCouncilof
Canada, Ottawa, Canada K1A 0R6, Max-PlanckInstitutfurExperimentelle Medizin,AbteilungImmunchemie, 37075 Gottingen, Germany, and Medizinische Hochschule Hannover, InstitutfurMedizinische Mikrobiologie, 30625 Hannover, Germany
ReceivedMay23, 1994; RevisedManuscriptReceivedNovember9, 1994®
abstract: The antigen binding fragment
from
an IgG2a k murine monoclonal antibodywith
specificityfor
a-(2—1•8)-linked sialic acid polymers has been prepared and crystallized in the absenceof
hapten.Crystals were grown by vapor
diffusion equilibrium with
16—18% polyethyleneglycol
4000 solutions.The structure was solved by molecular replacement methods and refined to a conventionalR factor
of
0.164
for
data to 2.8 A. The binding siteisobserved to display a shapeand distributionof
charges thatiscomplementarytothat
of
thepredicted conformationof
theoligosaccharideepitope.A
thermodynamicdescription
of
ligand binding has been compiledfor
oligosaccharides rangingin
lengthfrom
9 to 41residues, and the data
for
the largest ligand has been used in a novel way to estimate the sizeof
the antigenbinding site.A
modelof
antigenbinding is presentedthat satisfies thisthermodynamicdata, aswell
as a previously reported requirementof
conformational specificityof
the oligosaccharide.X-ray
crystallographic and thermodynamic evidence are consistent
with
a binding site that accommodates atleast eight sialic acid residues.
The capsular polysaccharides
of
group B meningococci and Escherichia coli capsular typeKl,
which both causehuman meningitis, consist
of
ahomopolymer ofa-(2—*8)-linkedsialic acid (V-acetylneuraminic acid;Jennings etal.,
1984). The polymer is known to be poorly immunogenic
(Wyleet al., 1972; Jennings & Lugowski, 1981). That this
poor immunogenicity is due to molecularmimicry is
sup-ported bytheidentificationofstructurally identical
a-(2—8)-linked sialooligosaccharidesinhumanandanimalfetaltissue (Finne et al., 1983). Chains
of
a-(2—8)-linkedsialic acidare developmental antigens (Rutishauser
&
Jessel, 1988a),are carried by glycoproteinsinvolvedinneuralcelladhesion
(Rutishauseretal., 1988b;Phillipsetal., 1990; Brandleyet
al., 1990; Corralet al., 1990; Berg et al., 1991), and have also beenidentifiedin othertissuesincludinghumantumors (Troy, 1992). Despite its poor antigenicity, antibodies to a-(2—1"8)-polysialicacidcan beproducedin special circum-stances. One suchmonoclonal antibody (IgG2a,designated
*
Coordinateshave been depositedwith theProtein Data Bank at
BrookhavenNational Laboratories,code 1PLG.
*
Correspondingauthors.
"
Present address:DepartmentofBiochemistry,UniversityofOttawa, 451 Smyth, Ottawa, CanadaK1H 8M5.
vNational ResearchCouncilofCanada.
§Present address:
DepartmentofChemistry, Carlsberg Laboratory,
Valby, Copenhagen DK-2500, Denmark.
aMax-PlanckInstitutfiir
Experimentelle Medizin.
11Medizinische Hochschule Hannover.
1Present address: Ontario Cancer Institute and
Department of Medical Biophysics, University of Toronto, 500 Sherboume St.,
Toronto, Ontario,Canada M4X 1K9.
* Present address:
University ofAlberta, DepartmentofChemistry, Edmonton, Alberta, CanadaT6G2G2.
8
Abstract published in Advance ACS Abstracts, February15, 1995.
mAb735)has beencharacterized (Froschetal., 1985), and its heavy andlightchains have been sequenced (Vaesen et al., 1991; Klebertet al., 1993).
This antibody, itsFab,1 andothera-(2—1"8)-polysialicacid specific antibodiesallrequirean unusuallylong segment
of
haptenforbinding, and theaffinityappears to increasewith
increasing chain length(Jennings et al., 1984, 1985; Finne
&
Makela, 1985;Rabatetal., 1988;Hayrinenetal., 1989).It has been proposed that antibodies recognize a specific
helical conformation
of
this polysaccharideandaminimumof
10 residues is required to form a helical epitope. Thevalidity
of
this propositionwas further strengthenedby theidentification of a common length-dependent epitope
re-sponsible forthe cross-reaction
of
a-(2—8)-polysialic acidand poly(A) with a human monoclonal macroglobulin
(IgMN0V; Rabat etal., 1988) and the known propensity
of
poly(A) to form helices of n = 8—10
(Yanthindra
&
Sundaralingam, 1976). Thisconformationaldependenceonchain lengthhas been confirmed by NMR (Michon et al., 1987), and similar chain length-dependent conformational epitopes have been described for other antigenic capsular polysaccharides (Wessels et al., 1987; Wessels & Rasper, 1989). Bymeans
of
potentialenergycalculationsandNMR, it was shown that, although mostly randomcoil in nature, a-(2—1-8)-polysialic acidcan adopt extended helicalconfor-mations where n ^ 9 (Brisson et al., 1992), the stability
1
Abbreviations:ELISA,enzyme-linked immunosorbentassay; Fab,
antigenbinding fragment; PEG,polyethyleneglycol;LI,L2, L3, HI,
H2, and H3,the six hypervariable loops as defined by Kabat et al.
(1991);HEPES,AL(2-hydroxyethyl)piperazine-Y-2-ethanesulfonicacid;
SDS—PAGE, sodium dodecyl sulfate-polyacrylamide gel
electro-phoresis.
0006-2960/95/0434-6737$09.00/0 Published 1995 by theAmerican Chemical Society
Downloaded by NATL RESEARCH COUNCIL CANADA at 11:57:01:390 on July 05, 2019
6738 Biochemistry, Vol. 34, No. 20, 1995 Evans et al.
ofwhichare dependenton itscarboxylategroups(Baumann etal., 1993). Ithas also beenshownthatmAb735 displays
afairlyhighdegreeofspecificityforthe 77-acetylneuraminic
acid polymer,asthereisonlyasmall cross-reactivity (%5%)
observed with the 77-propyl analogue (Jennings et al., unpublished results).
The determination
of
the three-dimensional structureof
theantigen binding fragmentandathermodynamic descrip-tion
of
the hapten bindingof
such a system are necessarysteps in extending our understanding
of
the interactions between antibodies andspecific conformationsof
antigenic polysaccharides. We now report the preparation,thermo-dynamic characterization, crystallization, molecular replace-ment solution, and2.8
A
resolution structureof
aFabfromthe murine monoclonal antibody mAb735 specific for
a-(2—8)-linked polysialicacid. Wealso present theresults
of
modeling studies on the natureof
the antibody-antigen complex carried out using the structureof
thebinding site reported herein and the structureof
the antigenpreviously reportedfromNMR experiments (Brisson etal., 1992).EXPERIMENTAL
PROCEDURESPreparation
of
Fab. TheFab was prepared by digesting30.4 mg
of
mAb735 (IgG2a) in 15 mLof
10mM Tris-HClcontaining 50 mM dithiothreitol (pH 8.0) with 152 pg
of
papain (IgG:papain = 200:1) for 5 h at room
temperature. The solutionwas dialyzed at4 °Cagainst three changes
of
10 mM Tris'HCl (pH 8.0) and applied to a DE52 ion
exchange column. TheFab fragmentwas eluted with the
same buffer. SDS—PAGE
of
thepurifiedFabunderreduc-ing conditions showed two bands
of
apparent molecular weight30and26kDa, correspondingto theFdandLchainsof
the Fab, andisoelectric focusing showedasingle bandatp/
8.4.Preparation
of
Oligomersof
Sialic Acid. Oligomersof
sialic acidwere obtained by partial acid hydrolysis
of
the sodium saltof
colominic acid purchased from Nacalai Tesque, Inc. (Kyoto,Japan); 100mgof
colominic acidwasdissolved in 3 mL
of
water, thepH brought down to 2.0, andthe sample heated to 80 °Cfor 1 h. Aftercooling,thepH was adjusted to 7.0, and crude fractions
of
varyingmolecular weight withchain lengths ranging from 9 to 15
were obtainedfrom BioGel P-10,
200-400
mesh (BioRad Laboratories) chromatography eluted with 30 mMam-monium bicarbonate buffer, pH 8.0. The fractions were
dialyzedagainst water andlyophilized. Furtherpurification was carried out by preparative exchange chromatography
(Dionex anion) using a Carbo-Pac PA1 (9 x 250 mm)
column with pulsed amperometric detection. The eluant
consistedof0.15 Msodium hydroxidewithagradientfrom
50 mM to 1 M sodium acetate (followed by post column
additionof0.5Msodium hydroxide). Afterpassagethrough
thedetector,theeluantfractionswere immediately adjusted
to pH7.0withacetic acid. Ahomogeneouspreparation
of
colominic acid corresponding to an approximately 12 kDa
polymer (equivalent to approximately 41 residues) was
obtained as the highest molecular weight fraction from
Sephadex G-100 chromatography
of
the sodium saltof
colominic acid using 10 mM sodium phosphate, 0.15 M
sodium chloride, pH 7.2.
Microcalorimetry. Thethermodynamicparametersforthe
binding of the a-(2—8)-linked sialic acid polymer by the
mAb735 Fab were obtained by titration calorimetry using
an OMEGA titrationmicrocalorimeterfrom MicroCalInc.,
(Northampton,
MA)
(Wisemanetal., 1989). Asolutionof
76 /<M Fab in 10pM
sodium potassium phosphatebuffercontaining0.15 M sodium chloride, pH7.4, was titratedat
25 °Cwithligandinthesame buffer. Twenty injectionsof
5
pL
froma100pL
syringe stirringat400 rpmwere carried out from a solutionof
13.35 mg/mL ligand. The thermo-dynamicsof
thebinding ofoligomers to mAb735 IgG wasdeterminedinthe same way. Dataanalysis
of
the binding isotherms was carried out as described previously (Sigur-skjoldet al., 1991).Crystallization andX-rayDiffractionAnalysis. Crystals
of
mAb735were obtained bythehanging drop method.The reservoir solutionwas composedof
16—18%PEG-4000,0.1M HEPES atpH 7.5, 0.1M (NH4)2S04, and 0.01 M NaN3. Thedrops were madeby mixing6.0
pL of
6.7 mg/mLFabwith6.0
pL of
reservoir solutionon thecoverslipandseeded1 day later with microcrystals grown previously. Crystals
appearedafter1 weekandgrew toan average size
of
0.3 x0.3 x 0.4 mm3. Crystals
of
mAb735are orthorhombic,witha = 79.14
k,b
= 91.21 A, andc = 141.4 A.Systematic
absencesshowed thespacegrouptobeeither1222oxI2\2\2\, withone molecule per asymmetric unit.
X-ray diffractiondatafromasinglecrystalwere collected
on aSanDiegomultiwiresystemtwo-detectorsetupmounted
on aRigaku RU-200 rotatinganodeX-raygenerator operated at 40kV and 120mA. The detectors were positionedat 6 values
of -12.0°
and30.0°, and thediffraction sphere wassampledin seven sweeps. A total
of
87625 measurementswere made
of
10590 unique reflections, outof
a totalof
10611 possible reflections, yielding a data set 99.8% complete to 2.8
A
resolution. No explicit corrections forabsorptionor decay were applied.
Molecular Replacement Solution. The structure was
solved in space group 1222 via molecular replacement
methods. Solutionproceededintwo stages, usingthe
anti-Brucellaabortus YsT9.1 Fab structure (Rose et al., 1993)
as amodel. First,theorientationandposition
of
theconstant regionwere determined using theMERLOTpackage (Fitzger-ald, 1988). Many combinationsof
resolution range and acutoffwere tried, with the cleanest results obtained using
datain the range 6.0—4.0
A
resolution with/ > 3o(l).The positioning
of
the variable region was less straight-forward,as no combinationof
data range anda cutoffusingMERLOT led to a solution. The correct position of this
fragmentwas locatedvia apackingserachutilizingthenow
correctly positionedconstantregion, by allowingthevariable regiontorotateabout the Fab ‘hinge’residuesin2°intervals,
whilemonitoring all intra-andintermolecularcontacts. This
procedure definedan allowed sectionofapproximately 15°
ofarc inwhichthevariable region couldbeaccommodated bythecrystal packing. However,asix-dimensionalfine grid search about the allowed section using the entire variable region
of
the YsT9.1 Fab model did not yielda match to the observed diffraction pattern. The final molecularre-placement solutionrequiredthe separate positioning ofthe heavy and light chains of the variable region
of
the Fab, usingtheprogramBRUTE(Fujinagaand Read, 1987),withtheentire constant regioninputas a stationary atoms set.
All
refinementswere carriedoutwiththeprogramXPLOR(Briingeret al., 1988)usingdataintheresolutionrange6.0—
Specific for a-(2-*8)-Polysialic Biochemistry, Vol. 34, No. 20, 1995 6739
Figure 1: Isotherm for the binding of the Fab to the 12 kDa a-(2—*8)-linked sialic acid oligomer. The dottedline shows the calculatedfit(Sigurskjoldetal., 1991) assuming that two Fabsbind
each 41-mer oligosaccharide.
/
theYsT9.1 model. Rigidbody refinement
of
themolecular replacement solutionof
the mAb735 began with just the variable and constant regions defined as rigid bodies and thenwitheachof
thosedivided into heavyandlightchains.Afteratotal
of
50cycles,theRfactor droppedto 0.33fromthe
initial
0.48. Refinementproceededwithseveralroundsof
simulated annealingwith aninitial
temperatureof
3500 K. Conventional positional refinement with an overall temperature factor applied further reduced the R factor to 0.26 after 50 cycles. After a single roundof
manualintervention, refinement proceeded again using an overall
temperature factor to yield an R factor
of
0.24 after anadditional50 cycles. To avoid biasingthe struture,no water molecules were included intherefinement.
At this point in refinement, the sequences for both the
heavyandlightchains becameknown (Vaesenet al., 1991; Klebertetal.,1993) andwere imposedon the structure. The
final set
of
refinements includedtheplacementof
41 watermolecules and individual atomic temperature factors were
allowedtorefine. Convergencewas reached afteratotal
of
50cycles toyieldafinalRfactor
of
0.164 ataresolutionof
2.8 A.
RESULTS
AND
DISCUSSIONThermodynamics. Three parameterscan beobtainedfrom
nonlinear regression analysis ofthe calorimetric isotherm: themolar enthalpy(AH°),thebindingconstant (K),and the apparent concentration
of
binding sites (Fab) (Wiseman etal., 1989; Sigurskjoldetal., 1991), and fromthese values,
AG° and AS° can be calculated. The isotherm for the bindingoftheFab to the 12kDaa-(2—8)-linkedsialic acid polymer consisting ofapproximately 41 residues is shown in Figure 1. (Unless otherwise stated, all figures were
generatedwiththeSETOR programpackage; Evans, 1993.) This isotherm is a smooth sigmoid curve, i.e., it describes
only a singlebinding constant. This means that
if
thereismore than one epitope on this polysaccharide chain, then
the binding
of
one Fab to the polysaccharide does notsignificantly facilitate (nor debilitate) the binding
of
asubsequent Fab. Such cooperativity in binding has been observed, e.g. , for the type
III
group B Streptococcuscapsularpolysaccharide(Wessels et al., 1987). Theisotherm
was fitted fourtimes assuming that thepolymercontained
either one, two, three, or four epitopes for binding to the Fab,respectively,and the apparentconcentration
of
Fabwasdeterminedineachfit. The regressionresultsare shownin
Table 1, andonly thevalue
of
73pM
Fab—corresponding to two epitopeson theligand—is consistentwiththeknown concentrationof
76 Fab. Itisconcluded thatapolymerof
approximately41 sialic acidresidues can accommodateonly two Fab molecules. The calorimetric measurements therefore provide an estimate
of
the extent to which anindividualFab renders adjacent segments
of
theantigen chainunavailableforbinding. Inthiscase,themaximum number
of
sialic acidresidues involvedis from 15 to 20, anumber thatis consistent with abinding site around 10residues inlength. This also demonstrates anovel use
of
calorimetryto determinethe size
of
abinding epitope. The associationisenthalpy drivenwithalargeunfavorablechangeinentropy. Strong enthalpic interactions combined with unfavorable entropychangesare usually observedfor
protein—carbohy-drateinteractionswithlongsugarchain lengths[Sigurskjold
et al. (1991) andreferences therein].
For comparisonthethermodynamics
of
thebindingof
the whole mAb735 IgGto oligomers from 9 to 15 residuesinlengthare shownin Table 2. Thebinding
of
these ligands (Ka ~ 4 x 104M-1) is 1 order of magnitude weakerthanforthelong chain (Ka ^ 3 x 105M-1)- Thereseems tobe
little
or no chain lengthdependence foranyof
the thermo-dynamic functions in this rangeof
chain lengths, whichseems to contrast with previous observations (Jennings et
al., 1984, 1985; Finne
&
Makela, 1985; Rabatetal., 1988; Hayrinenetal., 1989);however, this trend maybeconcealed by thelarge standard deviations observedfor —TAS°. Wecan observe a relatively strong chain length dependence
between binding
of
the long polymer and the shorter oligomers. Theenthalpyof
bindingforthelongpolymeris twice thatof
theshorter ligands. Thusit seems thattheFab is able to take full advantageof
the enthalpic interactionsfor the whole epitope
of
about 20 residues. The entropy change, on the otherhand, has becomeprogressively moreunfavorable, —TAS° being 3 times greaterfor thepolymer than for the shorter oligomers and thus compensating for
some
of
the enthalpy gain. This reflects that—as expec-ted—thereisastrongentropypenalty inbinding longerchains(if
—TAS° for the polymer were just twice that of the oligomers, K wouldbeabout 109M_l).
X-ray Crystal Structure Analysis. A summary of the
deviations between stereochemicalparametersderivedfrom
the final cycle of refinement with ideal values (Table 3)
shows that this structure displays geometry thatis in good
agreement with literature values and has all non-glycine residues,withtheexception
of
V56on the lightchain,inor near theallowed regionsofaRamachandran plot, Figure2.Additionally,examinationover thecourse ofthepolypeptide
chain shows that, with a few exceptions distant from the bindingsite,all main chainand sidechain atomsare observed to occupy appropriate electron density. Figure 3 is an
electron densitymapoftheheavychain
HI,
which illustratesthe well-defined density that is typical for residues in the
6740 Biochemistry, Vol. 34, No. 20, 1995 Evans et al.
Table 1: ThermodynamicsandDeterminationofEpitopeSizeoftheBindingby mAb735 Fabofa 12kDa Polymerof a-(2—8)-Linked Sialic
Acid(CorrespondingtoApproximately 41 Residues)Obtained byTitrationMicrocalorimetry0
assumedno. ofepitopes K (xlO-5M-1) AG° (kJM"1) AH° (kJM-1) —7AS° (kJM-1) apparentcone ofFab(«M) 1 7.3±4.0* —33.8±1.4 — 161.6± 15.1 127.8± 15.8 37±4 2 3.4± 2.0 —31.9±1.5 -81.6±8.5 49.8±9.3 73± 8 3 2.4± 1.4 —31.0± 1.4 -54.1 ±5.3 23.1 ±6.1 110± 13 4 1.7±0.9 -30.1 ± 1.4 -41.0 ±4.1 10.9±4.9 146± 17 0The concentrationof
mAb735Fabintheexperimentwas 76wM.bUncertainties
represent three standarddeviations.
Table2: Chain Length DependenceoftheBinding of
a-(2—8)-Linked SialicAcidOligomerstomAb735 IgG Obtained
byTitration Microcalorimetry chain K AG° AH° -TAS° length (x10-4M_l) (kJM"1) (kJM-1) (klM”1) 9 3.9± 0.9° -26.2 ± 0.6 -45.4 ± 6.7 19.1± 6.8 11 4.8± 0.9 -26.7± 0.5 -42.6 ± 2.9 15.8± 2.9 14 5.9± 2.9 -27.2 ± 1.2 -38.7± 2.8 11.5± 3.1 15 4.4± 1.9 -26.5 ± 1.1 -41.6 ± 6.8 15.9± 7.3 average 4.4± 0.9 -26.6±0.5 -41.0 ± 2.6 14.4± 2.7 “Uncertainties
represent three standard deviations.
Table3: StereochemistryoftheFabfrommAb735 after Refinementto 2.8AResolution
stereochemical rms deviation refinement restraint refinement parameter fromidealvalues weighting parameter
bonddistances(A) 0.014 0.030
angledistances(A) 0.052 0.040 planar1—4distances(A) 0.065 0.060
planes(A) 0.022 0.030
chiral volumes(A3) 0.125 0.125
cudihedralangles(deg) 3.5 15.0
single dihedral contacts(A) 0.232 0.250
multipledihedralcontacts (A) 0.201 0.250
-120 - ^
o
-180 MI
ytt
U--U-J-,-i
-180 -120 -60 0 60 120 180
9
Figure 2: Plotofmain chain dihedralangles (Ramakrishnan &
Ramachandran, 1965)ofall non-glycine residuesofmAb735. Theframework regionsofthemolecule adopt conforma-tions similar to those reported for other Fabs and display theusualimmunoglobulinfold withan elbowanglebetween thepseudo-2-foldaxes
of
thevariableandconstant domaindimersequal to 166°. Themoleculesare packedinthe usual head-to-tail fashion observed formost Fabs, withthe long axis
of
the molecule nearly coincident with the unit cellc-axis. Seven
of
the 15 observed intermolecular contactsare betweenresiduesinthehypervariable regionsand the C domain dimers
of
adjacentmolecules, mostlyinvolvingtheLI
andHI loops.Astheshapeand chargedistribution
of
theoligosaccharide are of paramount importance in describing its antigenicproperties, the binding surface
of
the Fab is also best described in that manner. The antigen binding surface is bimodal in thatit
undergoesastrikingreversalinshape andcharge distribution along the interface between the heavy
andlightchains. Theleft side
of
thesite(asinFigure4) iswhere L3, H2, and the ‘tip’
of
H3 are observedto form aconvex surfacewhichdisplays threepositivelychargedlysine residues,Lys H59,LysH65, andLys H101. Therightside
of
the site finds L2 and thebulkof
H3 meetingto form agroovewhichhas arelativelyunchargedsurfaceexceptfor
thenegatively chargedAsp
HI05
residue at thebottomof
thegroove, with positively charged Arg H98just above
it
on the side. Loops
LI
andHI display no charged residuesexcept Asp H31, which is well away from the observed groove.
ModelingtheHapten into theBindingSite. X-ray crystal-lographicstudiesofcomplexes
of
Fabswitholigosaccharideshave shown that the specificity
of
the binding site on the Fabsusuallyencompassesone tofoursugar residues (Cygler etal. 1991;Oomenetal., 1991;Vyasetal., 1993);however, thereare severalreasons to suppose thatthebinding siteonmAb735 is much larger. Firstly, binding studies have demonstrated that a critical chain length
of
10a-(2—8)-linkedsialic acidresidues isrequiredinorderforthe hapten
tobindtomAb735 (Hayrinenet al., 1989), andthishasbeen confirmedinour thermodynamicanalysis. Secondly,NMR
studies in conjunction with potential energy calculations (Brissonetal., 1992) are consistentwiththe presence
of
anextendedhelicalepitopein a-(2—8)-polysialicacid. Thirdly,
from examples where binding studies have been reported (Jennings etal., 1985; Finne
&
Makela, 1985;Kabatetal.,1988; Hayrinen et al., 1989), all antibodies specific for
a-(2—8)-polysialic acid reported to date are based on an
extendedhelical epitope. Nonehave been detected with a
specificityforshortera-(2~*8)-sialooligosaccharides. Thus, anymodelofhaptenboundin the binding site of mAb735
must reflectthis helicalnature ofthe antigenicformofthe
hapten andallowcontactbetweenantibodyand antigen along
aspanthatis
of
theorderof
10residuesinlength. Finally,themodel must explainthespecificity thatmAb735 shows
fortheA-acetyl-vs N-propylneuraminicacid. Althoughthe
'H
NMR analysis provides overwhelming evidence for the requirement of a helical conformation for the antigen, theSpecific for a-(2—8)-Polysialic Biochemistry,
Figure 3: Stereoview of the tip of the H3 loop (thick lines) and surrounding residues, showing the corresponding observed electron density (thin lines) plottedat a 1.5<jlevel. With fewexceptions, all residues in the protein display excellent agreement withobserved electron density.
Figure4: Stereoviewoftheantigenbindingsurface(Connolly, 1981)ofmAb735 looking down thepseudo-2-foldaxis. Those portions
ofthesurface due to charged atoms are shown withdenser dotdistributionand havetheirparent residues labeled.
Figure5: Side and top views withrespect to the helixaxis for helical modelsofa-(2—8)-linked sialicacid 10-mers constructed
to fit 'H NMR solutiondata(Brisson et
al„
1992). Although thepitchofthe helixthatcan beaccommodated from the NMRdata
can vary significantly, all helix models require the negatively chargedcarboxylicacid groupsto facetheinteriorofthehelixwhile
the N-acetyl groups face the exterior ofthe helix. The index n
showsthenumberofresidues perturn foreachhelix.
exact pitchofthehelix thatcan begeneratedto fit thisdata
can vary significantly. Figure 5.
Flelicesare definedin terms ofaset
of
helical parameters (n,h) with n being the number of residues per turn and hbeing the translation along the helix axis ofcorresponding residues. The pitch is defined as n x h. Helices were
generated as described before (Brisson et
al„
1992). The rise per residue, h, forthe n= 6helix was 6.0A,and5.8Afor the ;i = 9 helix. Extended helices
of
nearly identical energy can be generated forn = 6—11
by a slightchange
(30°) in only one
of
the four dihedral angles linking two residues (Brisson et al., 1992). For the n = 6and n = 9helix, this dihedralanglediffers by only 20°. This demon-strates that the pitch
of
these extended helices can varygreatly by minimal rotation about only one bond and with
very little change in energy.
All
of
thesehelicalmodelsof
polysialic acidhave several importantfeatures in common. Mostsignificantly, in eachmodel the negatively charged carboxylic acid groups face
the interior of the helix with a repeat distance
of
5—8 Awhilethe (V-acetylgroupsface theexteriorofthehelix. This
means thatforany
of
themodels,over one turn ofthehelix,one endoftheoligomer wouldpresent thenegativelycharged
carboxylic acid groups to the binding site on the antibody
while the midpoint of the oligomer would present the
relatively uncharged but protruding/V-acetyl groups. This reversal
of
environment fromone endofthe epitope to the other is themirrortothat described abovefortheFabbinding site, whereone extremeof
thebindingsite presents aseriesofpositivelychargedlysine groupswith thecorrect spacing
to form saltbridges with the carboxylicacid groups on the antigen, whiletheotherextreme ofthebindingsite presents
6742 Biochemistry, Vol. 34, No. 20, 1995 Evans et al.
Table4: SuccessiveRotationofSialicAcidResiduesinthe Model ofBound Hapten
sialic acidresidues rotation (deg)
2-3 66.5 3-4 58.0 4-5 74.5 5-6 50.3 6-7 72.3 7-8 68.2 8-9 52.6
agroove whichcan accommodate andrecognizethe IV-acetyl groups.
The observation
of
shape and charge complementaritybetween antibodyandantigenstronglysuggestsamodelfor
mAb735—polysialic acid binding, and unlike the helix
modelsderivedfromthe
NMR
data, thebinding siteon the Fabhasvery definite dimensionsandcan onlyaccommodatea.narrow range
of
helixpitch. In order for the hapten toformboththesalt bridges and beproperly oriented to project theIV-acetylgroupsintotheobserved grooveon the surface
of
the Fab, the oligosaccharide helix must makeap-proximately V2
of
aturn within 17 A. Thiscorresponds tohelixa
of
Figure5,withn = 6residuesperturn andapitch
of
n x h = 36A.The approach taken for modeling the complex was
conservative. No main chain atoms
of
the protein weremoved. Sidechain conformations forboththeproteinand theoligosaccharidewere allowedto vary betweensterically favoredrotamers. Inorderto bestcomparethe
fit
between oligosaccharide model and Fab, the basic helical shapeof
the searchmodelwas preserved, andTable4 showsthatthe angleof
rotationbetween each successivepairof
sugarrings is close to the 60° expectedforan n = 6helix. However,small changes inthedihedral angles ofthelinkingregions
of
the oligosaccharide were required to bring suitable hydrogen-bondingpartnersinto proximity. TheNMRdatawill
accommodate a rangeof
pitch for the oligosacchride links, andalthough themodelsthatare presentedinFigure 5containlinksof
identical pitch,anymodelcan containlinksof
different pitch. Two water molecules that had beenlocated inthebinding siteduringtherefinement procedure
were movedlessthan 1.0
A
suchthattheoriginal hydrogen-bonding partner was maintained. A third water molecule,Table5: NumberandTypeofHydrogen Bonds Foundon
OligosaccharideResiduesintheModel
sialic acidresidue total toCOO toA-acetyl
2 3 0 3 3 7 0 2 4 5 0 3 5 4 2 1 6 4 1° 0 7 2 1 0 8 2 1“ 1 9 2 1“ 0 “
Chargedresidueinteraction.
Wat-21, whichwas inthe secondhydrationshell, had tobe
removed tomakeway forthe oligosaccharide.
The
initial fit
of
the 10-merhelix in Figure 5ais showninFigure6. Thethreepredictedsaltbridgeswiththelysine residuesareeasily formedwithsmallchangesinthedihedral angles
of
thelinking regionsof
the oligomer, andthe turnof
the helix is adequate to bring an appropriate A-acetylgroup as well as much
of
its parent sugar residueinto the groove observedon thefaceof
the Fab. The highdegreeof
complementarityofthe
initial
fit of
thehelixto the binding surfaceof
theFabis evident,aseachone ofeight consecutiveresidues of the oligosaccharide makes significant overall contact with the protein, forming at least two hydrogen bonds,Table5andFigure7. Table5alsoreflectsthehelical nature
of
theantigen, wherethefirstresiduesin the haptenare bound chiefly through the A-acetyl groups, while the latterresidues form saltbridges to the carboxylate anions.
Withtheexception
of
LI,
eachof
thehypervariable loopscan make at least one hydrogen bond directly with the modeled hapten; however, residues on LI can serve to stabilize the position of a water molecule in a position to
formastrongbridgeto the hapten,Figure7. Nosignificant contact is madebetweentheframeworkresidues oftheFab and the oligosaccharide model. Table5 andFigure 7 both showthatthemodel
of
the antigenic formof
the oligosac-charide can make as many as 30 hydrogen bonds withthe Fab,withaAH°of
about —41kJM-1. Thiscompareswell with the crystal structure and thermodynamics (Bundle et al., 1994) of theSalmonella O-polysaccharide antigen Fab complex (Cygler et al., 1991), which shows 12 hydrogenbonds with aAH°
of
about —20 kJ M_1.
Figure 6: Stereoviewofthe fit ofthehelical modelofa-(2—8)-linked sialic acid to thebinding surfaceofFabmAb735. The Fab is
viewed approximately downitspseudo-2-foldaxis,withthelightchainpositionedabove and the heavy chainbelow the hapten. Theleft
sideofthediagramshows the salt bridgesformedbetween three surfacelysinegroups and thecorrespondingcarboxylicacid groupson the hapten. Theright sideshowsthe contact oftheA-acetyl groupsofthe haptenwith surfacefeatureson the Fab.
A
FabSpecific for a-(2—8)-Polysialic acid Biochemistry, Vol. 34, No. 20, 1995 6743Figure7: Stereoview ofthespecificcontacts madebyan eight-residuesegmentoftheoligosaccharideshownin Figure5ato the surface
ofthe Fab,showingthehighdegreeofcomplementarilyofshapeand charge between theprotein and carbohydrate. Modeled hydrogen
bondsare shownwithdashedlines. Theproteinresiduesare labeledwiththeir one-letter amino acidcode and toindicatetheirparentlight (L)or heavy (H) chain. The oligosaccharideresiduesrun fromSia-2(top) toSia-9 (bottom).
The greatestnumber
of
interactions between proteinand hapten are found alongthe surfacegrooveon therightsideof
theFabas seen inFigure4 andinvolvingSia-2—Sia-5. Itisin thisarea thatthespecificity
of
mAb735for A-acetylvsA-propyl polysialic acid becomes apparent, as a series
of
hydrogen bonds is formed between several sialic acid residues and thebottom and sides
of
the groove thatwouldnot bepossibleforthelongerA-propylgroup. OneA-acetyl group (on Sia-3) forms a hydrogen bond with Asp H105.
ResiduesSia-6—Sia-9are primarilyinvolved through their
respective carboxylic acid groups, which take part in one
hydrogen bond, and threesalt bridgeswith lysine residues
on the protein. There are at least six additional hydrogen bonds thatcan bereadily modeled betweentheprotein and other atoms
of
the sugar, butthe formationof
salt bridges between theprotein and thecarboxylic acid groups of the sialooligosaccharidesisan important featureintherecogni-tion
of
thehelix,and compareswell withthecrystalstructureof
thecomplexof
asialic acidanaloguewithinfluenza virusneuraminidase (Vargheseetal., 1992). Itisin thisarea that theslight cross-reactivityofmAb735withtheA-acetyland
A-propylneuraminic acid couldreside.
Althoughthe chargedistributionon theproteinis the most
significant feature
of
the left sideof
Figure 4, a role canalsobe envisioned fortheright side, wherethe positively-charged Arg L55 is interacting through a bridging water molecule with the carboxylic acid group on Sia-4. The negative chargeon Asp
HI05,
locatedatthebottomof the groove where it can form ahydrogen bond to an A-acetylgroup, mayalsoserve to directtheorientation
of
the haptenby repulsing its carboxylic acid groups.
In all, theextendedhelical model proposed by Brissonet al. (1992) with n = 6residues
per turn andpitchofn x h
= 36 A has a
shape and charge distribution that is complementarytotheFabfrommAb735,andcan beshown
to form a significantnumber
of
specific contacts with theFabover aspan at least8sialic acidresiduesinlength. This
is close to the 10 residues predicted from earlier binding studies (Hayrinen et al., 1989) and is confirmed by our
thermodynamic analysis.
CONCLUSIONS
X-ray crystallographic, thermodynamic, and NMR
evi-dence are consistent in the prediction
of
a binding site onmAb735 which is at least eight residues in length. The
crystallographic study is also consistent with the earlier prediction
of
a helical nature forthe antigenic formof
the oligosaccharide and provides a molecular basis for the recognitionof
the A-acetylneuraminic acid. Trials areunderway to grow crystals
of
the Fab complexed witholigosaccharides
of
different lengths; however, no crystalshave beenobtained to date.
REFERENCES
Baumann,H.,Brisson, J.-R.,Michon,F., Pon,R.,&Jennings, H.
J.(1993)Biochemistry32,4007-4013.
Berg, E. L., Robinson, M. K., Mansson, O., Butcher, E. C., & Magnani, J. L. (1991)J.Mol. Biol. 266, 14869-14872. Brandley,B. K., Swiedler, S. J.„ & Robbins, P. W. (1990) Cell
63, 861-863.
Brisson,J.-R., Baumann,H., Imberty, A., Perez, S.,& Jennings, H.J. (1992) Biochemistry31, 4996-5004.
Briinger, A.T. (188)J.Mol. Biol.203, 803-816.
Connolly, M. L. (1981)Quantum Chem.ProgramExchangeBull.
I,
75.Corral,L., Singer, M. S.,Macher, B. A., & Rosen, S. D. (1990) Biochem.Biophys.Res. Commun. 172, 1349—1356.
6744 Biochemistry, Vol. 34, No. 20, 1995 Evans et al. Cygler, M.,Rose, D.R.,&Bundle, D. R. (1991) Science 253,442—
445.
Evans, S. V. (1993)J.Mol. Graph. 11, 134-138.
Finne,J.,&Makela,P.H. (1985)J.Biol. Chem.260,1265-1270. Fitzgerald, P. M. D. (1988)J.Appl. Crystallogr.21, 273—278. Frosch, M.,Gorgen, I., Boulnois,G.J.,Timmis, K. N., &
Bitter-Suermann, D. (1985) Proc. Natl.Acad. Sci. U.SA. 82, 1194—
1198.
Fujinaga, M., & Read, R. (1987) J.Appl. Crystallogr. 20, 517—
521.
Hayrinen,J.,Bitter-Suermann,D.,&Finne,J.(1989)Mol.Immunol. 26, 523-529.
Jennings, H. J., & Lugowski,C.(1981)J.Immunol. 127, 1011—
1018.
Jennings, H. J.,Katzenellenbogen,E., Lugowski,C„Michon,F„
Roy,R., & Kasper, D. L. (1984) Pure Appl. Chem. 56, 893—
905.
Jennings, H. J., Roy, R., & Michon,F. (1985)J. Immunol. 134,
2651-2657.
Rabat, E. A. (1960)J.Immunol. 84, 82-85.
Rabat, E. A.,Lino, J., Osserman, F., Gamian,A., Michon, F.,& Jennings, H. J.(1988)J. Exp. Med. 168,699-711.
Rabat, E.A., Wu,T. T., Perry, H.M.,Gottesman,K.S.,&Foeller,
C. (1991)Sequencesofproteinsofimmunologicalinterest, 5th ed., National InstitutesofHealth,Bethesda, MD.
Klebert, S.,Kratzin, H. D., Zimmerman, B., Vaesen,M.,Frosch,
M., Weisgerber, C., Bitter-Suermann, D., & Hilschmann, N. (1993) Biol. Chem. Hoppe-Seyler374,993-1000.
Michon,F.,Brisson, J.-R.,&Jennings, H.J.(1987) Biochemistry
26, 8399-8405.
Oomen, R., Young, N. M.,& Bundle, D.R. (1991)ProteinEng. 4, 427-433.
Phillips, M. L., Nudelmann,E., Gaeta,F.C.A.,Perez,M., Singal, A.
K,
Hakomori, S.-I., & Paulson, J. C. (1990) Science 250,1130-1135.
Ramakrishnan, C., & Ramachandran, G.N. (1965)Biophys. J. 5,
909-933.
Rose, D. R.,Przybylska, M., To,R.J.,Kayden,C. S.,Oomen, R.
P.,Vorberg,E.,Young,N.M.,&Bundle,D. R.(1993) Protein
Sci. 2, 1106-1113.
Rutishauser,U.,&Jessel,T.M. (1988a) Phys.Rev.68, 819—857. Rutishauser, U., Acheson, A., Hall, A.
K,
Mann, D. M., &Sunshine, J.(1988b) Science 240, 53—57.
Sigurskjold, B. W., Altman, E., & Bundle, D. R. (1991) Eur. J. Biochem. 197, 239—246.
Troy, F.A. (1992) Glycobiology2, 5-23.
Vaesen, M.,Frosch, M., Weisgerber,C., Eckart, K., Kratzin, H., Bitter-Suermann, D., & Hilschmann, N. (1991) Biol. Chem.
Hoppe-Seyler372,451-453.
Varghese,J. N.,McKimm-Breschkin,J.L., Caldwell,J.B.,Kortt, A. A., & Colman, P. M. (1992) Proteins'. Struct., Fund.,and
Genet. 14, 327-332.
Vyas,M. N., Vyas, N. K., Meikle,P. J.,Sinnott, B., Pinto, B.M., Bundle,D.R„ &Quiocho,F.A. (1993)J.Mol.Biol.231,133—
136.
Wessels,M. R„ &Kasper, D.L. (1989)J. Exp. Med. 169, 2121—
2131.
Wessels, M. R.,Munoz, A., & Kasper, D. L. (1987)Proc. Natl.
Acad. Sci. U.SA. 84, 9170-9174.
Wiseman,T„Williston,S.,Brandts,J.F.,&Lin, L.-N.(1989)Anal.
Biochem. 179, 131—
137.
Wyle, F. A., Artenstein, M. S., Brandt, B. L., Tramont, D. L.,
Kasper, D. L., Altieri, P., Berman, S. L., & Lowenthal, J. P.
(1972)J.Infect. Dis. 126,514-522.
Yathindra, N.,&Sundaralingam,M. (1976) Nucleic AcidsRes.3,
729-747.